In a dimly lit youth centre in Sundsvall, a teenager is asked to make a choice. The prompt is simple, almost ordinary. Someone offers an opportunity, a short path to belonging, money, identity. The teenager hesitates, then selects an option on a screen. The story continues, reshaping itself around that decision. Consequences unfold.
This is not a game, although it borrows the language of one. It is an artificial intelligence designed to simulate the logic of gang recruitment, a digital mirror held up to a fragile moment in a young person’s life. Developed by Sundsvall municipality, the tool, titled “Med livet som insats”, allows young people to explore different paths and witness how quickly circumstances can change.
“It’s an innovative conversational tool,” said Robert Gatugård, who coordinates drug prevention work in the municipality, describing how the system opens up discussions about recruitment, drugs, online vulnerability and future choices.
Across Sweden, and indeed much of Europe, the urgency behind such experiments is clear. Gang networks are recruiting children at increasingly younger ages, a development often described as stark and without easy solutions. What distinguishes the work in Sundsvall is its focus not on enforcement, but on conversation. Here, AI is not a surveillance instrument but a narrative partner.
The structure is deceptively simple. Guided by a trained facilitator, a group of young participants navigates scenarios grounded in everyday life. Each step offers several choices, and the system adapts, weaving each response into a branching chain of consequences. At the end, the group reflects together. There is no winning or losing, only understanding.
“AI can help open up conversations that might otherwise be difficult,” Gatugård noted in a separate statement, emphasising that the technology is not meant to replace human interaction but to deepen it. This emphasis on augmentation rather than substitution recurs in Sweden’s broader approach. At the Royal Institute of Technology (KTH) in Stockholm, researchers are working on a different kind of system. Instead of simulating conversations with young people, their project aims to connect the adults around them.
The KTH initiative centres on an AI-driven “super brain”, a knowledge platform designed to gather insights from police reports, schools, social services, and civil society organisations. The goal is to make fragmented knowledge usable, and to strengthen coordination between actors who often operate in parallel rather than together.
“Getting different agencies to collaborate and coordinate activities effectively is a major challenge today,” said Susanne Nilsson, who leads the project. “There is a wealth of knowledge… but it is scattered among various actors.”
In this sense, the two approaches form a quiet dialogue. In Sundsvall, AI enters the room where young people sit, offering a space to rehearse decisions before they are lived. At KTH, AI works behind the scenes, attempting to align institutions so that the right support reaches the right person at the right moment.
Both efforts are built on the same underlying premise: that prevention depends on timing, and that timing depends on understanding.
Another voice from Sundsvall captures this shift in thinking. “Safety is not only about having more adults in the right place,” said Lisa Tynnemark, deputy chair of the municipal executive board. “It is also about finding new ways to reach young people in time.”.
There is something almost paradoxical in using machines to address what is ultimately a human question: why one young person chooses one path and another does not. Yet the designers of these systems seem acutely aware of that tension. The AI is constrained, guided, contextualised. It does not decide. It prompts.
The ethical considerations are never far away. Privacy, data protection, and the risk of over-reliance on automated systems remain central concerns, particularly in the KTH project where sensitive information from multiple actors is involved. But within those constraints lies a sense of cautious ambition.
What emerges is not a grand technological solution, but a series of small, deliberate interventions. A conversation that begins where silence might otherwise prevail. A dataset that reveals a pattern no single institution could see alone. A moment of hesitation, extended just long enough to allow reflection.
If there is a future being sketched here, it is not one in which AI replaces human judgement, but one in which it quietly reshapes the conditions under which that judgement is made. The screen in the youth centre does not know the teenager in front of it. It cannot feel the weight of their circumstances. But it can show them a story, and invite them to pause.
And in that pause, perhaps, lies the beginning of something else.
Photo: Sundsvall Municipality
Sundsvalls kommun i framkant med AI-stöd för svåra samtal | Sundsvalls kommun
